The Obstructive Sleep Disorders–6 survey.
Distribution of Obstructive Sleep Disorders–6 survey item responses. Higher scores indicate poorer quality of life (N = 101).
Customize your JAMA Network experience by selecting one or more topics from the list below.
de Serres LM, Derkay C, Sie K, et al. Impact of Adenotonsillectomy on Quality of Life in Children With Obstructive Sleep Disorders. Arch Otolaryngol Head Neck Surg. 2002;128(5):489–496. doi:10.1001/archotol.128.5.489
To determine the impact of adenotonsillectomy on quality of life (QOL) in children with obstructive sleep disorders (OSDs) before and after surgery.
Prospective, observational, before-and-after trial.
Seven tertiary pediatric otolaryngology practices.
Convenience sample of 101 children (mean age, 6.2 years) with adenotonsillar hypertrophy and OSD scheduled for adenotonsillectomy.
Adenotonsillectomy was performed in children for OSDs. Quality of life was assessed using the Obstructive Sleep Disorders–6 survey, a validated instrument for detecting QOL change in children with OSDs. Surveys were completed at the initial office visit (visit 1), the day of surgery (visit 2), and at the postoperative office visit (visit 3). Physical characteristics were assessed using tonsillar and orocraniofacial scales (visit 1). Satisfaction with health care decisions was assessed using the Satisfaction With Decision and Satisfaction With Office Visit scales (visit 1).
Main Outcome Measures
Short-term changes in QOL before (visits 1 and 2) and after (visits 2 and 3) surgery.
Changes in QOL before surgery were trivial or small, and smaller than changes after surgery (mean change score, 0.18 vs 2.3; P<.001). Large, moderate, and small improvements in QOL were seen in 74.5%, 6.1%, and 7.1% of children, respectively. Sleep disturbance, caregiver concern, and physical suffering were the most improved domains, although significant changes also occurred for speech and swallowing problems, emotional disturbance, and activity limitations. Five percent of children had poorer QOL after surgery, but no predictive factors were identified.
Adenotonsillectomy produces large improvements in at least short-term QOL in most children with OSDs.
THE TERM obstructive sleep disorders (OSDs) refers to the spectrum of sleep-disordered breathing that is severe enough to cause clinical symptoms. This includes children with upper airway resistance syndrome (UARS), in which the respiratory distress index is often normal on standard polysomnographic testing, and children with obstructive sleep apnea. The prevalence of sleep-disordered breathing in children is not exactly known but may approach 11%.1 The impact of OSDs on child functioning has been extensively documented. Children with OSDs may manifest an increase in total sleep time, nonspecific behavioral difficulties, hyperactivity, irritability, bed-wetting, and morning headaches. A more severe manifestation is failure to thrive, and, in general, untreated children are at risk for cardiovascular complications. Daytime sleepiness and obese body habitus, common features in adult sleep apnea, are frequently absent. The most common cause of OSDs in children is adenotonsillar hypertrophy, and adenotonsillectomy is a curative procedure in most cases.2-5
Anecdotal evidence for the success of this procedure in improving patient symptoms, health, and well-being is abundant; however, there have been no studies to date, to our knowledge, that document improvement in health-related quality of life (QOL) after adenotonsillectomy. In the context of OSDs, QOL describes the net consequences of sleep-disordered breathing on the child's daily activities, physical symptoms, social interactions, and emotional well-being. The effect of adenotonsillectomy on these areas of functioning is important information for parents considering the procedure for their child and is evidence of the procedure's effectiveness that can be presented to third-party payers in support of decisions for surgery.
A 6-item, disease-specific QOL instrument was recently validated as a measure for assessing change in patients with OSDs who undergo adenotonsillectomy.6 Data from the validation study suggested large improvements in QOL for most children undergoing the procedure. This study was designed to document the improvements in QOL using the validated instrument in a larger number of children from geographically diverse regions of the United States.
This study was conducted at 7 tertiary care pediatric otolaryngology practices across the United States. The study protocol received approval from the institutional review board at each site. Informed consent for enrollment in the study was obtained from each caregiver.
The study protocol was modeled after a recent study7 that examined QOL improvement after tympanostomy tube placement. Eligible participants comprised a convenience sample of caregivers of patients aged 2 to 12 years undergoing adenotonsillectomy for OSDs. Participants were required to be English speaking to avoid semantic issues in interpretation of the domains and symptom clusters.8 Caregivers were excluded if their child had other adenotonsillar pathologic findings or if another procedure was to be performed on the same day. Demographic data (age and sex) and tonsillar size were recorded for all patients, including those whose caregivers refused enrollment or dropped out of the study. Children were diagnosed as having OSDs by the attending pediatric otolaryngologist based on findings from a combination of any of the following modalities, as determined by the standard clinical practice of the attending physician: history, physical examination, lateral neck radiograph, nasopharyngoscopy, sleep audiotape, and polysomnographic study.
Data were collected at 3 times during the perioperative care of the child: at the initial visit (visit 1), at the day of surgery (visit 2), and at the postoperative visit (visit 3).
Before surgery, the physician completed a physical assessment form that documented the method of diagnosis (history, physical examination, nasopharyngoscopy, lateral radiograph, sleep audiotape, or sleep study), OSD symptoms (snoring, gasping, restless sleep, apneas, night terrors, daytime somnolence, daytime hyperactivity, enuresis, nocturnal sweating, and morning headaches), and physical examination findings, including degree of tonsillar obstruction (absent, 0%-25%, 26%-50%, 51%-75%, and 76%-100%), and a clinical orocraniofacial (OCF) scale.9 High scores on the OCF scale correlate with a diagnosis of sleep-disordered breathing (UARS and obstructive sleep apnea).9 The OCF characteristics included chin size (0-3, wide to small and triangular), steepness of mandibular plane (0-3, horizontal to steep), position of maxilla vs mandible (0-4, prognathic to significant retrognathia), height of hard palate (0-2, low placed to high placed), shape of face (0-3, square to long), length of soft palate (0-2, short to long), and intermolar width (0-2, wide to narrow).
Changes in QOL were determined by administration of the 6-item, health-related instrument for OSD, the Obstructive Sleep Disorders–6 survey (OSD-6), a validated survey for assessing health-related QOL in OSDs7 (Figure 1). The OSD-6 is composed of 6 domains that reflect functioning of the child regarding (1) physical suffering, (2) sleep disturbance, (3) speech and swallowing difficulties, (4) emotional distress, (5) activity limitations, and (6) level of concern of the caregiver relating to the patient's sleep disorder and associated symptoms. Each domain is represented by a question designed to reflect the global impact of an OSD-related symptom cluster on an individual child (Figure 1). Caregivers rated the domains on a scale from 0 ("no problem") to 6 ("could not be worse") based on how they felt the symptoms affected their child. The OSD-6 was given to the patient's caregiver at the 3 visits. The timing of visits was determined by the treating physicians, but it was requested that the final OSD-6 be administered 4 to 5 weeks after surgery so that the potential for sampling any symptoms of recovery would be bypassed.
To distinguish QOL changes resulting from surgery from potentially confounding factors, parents completed a satisfaction survey at visit 1. This enabled determination of whether changes in survey scores were independent of parental satisfaction with the office visit and physician interaction and the decision to have surgery. Validated surveys were abbreviated as described previously.7 In addition, patients also served as their own control as QOL changes were compared in individual patients before (between visits 1 and 2) and after (between visits 2 and 3) surgery, which also served to eliminate confounding.
The mean survey score was calculated by summing the individual domain scores and then dividing by 6 the total number of domains. A lower survey score indicates a better QOL. The change score was calculated by subtracting the postintervention OSD-6 score from the preintervention OSD-6 score. The change score was then used to define the level of change in QOL as trivial, small, moderate, or large and was summarized using the mean value and the 95% confidence intervals (CIs). The magnitude of clinical change for the mean OSD-6 score was classified as trivial (<0.5), small (0.5-0.9), moderate (1.0-1.4), or large (≥1.5) according to standard definitions for a survey on a 7-point scale.10
A secondary outcome measure, the standardized response mean (SRM), was also calculated. This statistic allows comparisons to be made between QOL instruments and was calculated by dividing the mean OSD-6 change score by its SD.11 An SRM of 0.2 reflects a small responsiveness to clinical change; 0.5, moderate responsiveness; and 0.8 or more, large responsiveness. The SRM was calculated for the mean survey score and the individual domains of the survey before and after surgery.
A matched-pairs design allowed each child to act as his or her own control. Preoperative and postoperative change scores are expressed as means with 95% CIs for each domain and for the survey as a whole. Preoperative and postoperative SRMs are also expressed with 95% CIs. The paired t test was used to compare the differences between preoperative and postoperative change scores and SRMs. Standard contingency χ2 analysis was used for categorical data. Multiple regression techniques were used to discern if any factors correlated with improvement in QOL. This technique was also used to explore which factors, if any, may have been associated with a poor QOL at baseline. All data were analyzed using SAS statistical software (SAS Institute Inc, Cary, NC). The cutoff level for statistical significance is P<.05. A sample size calculation showed that 32 participants would be required to detect a difference in change scores of 0.5 between the surgical and nonsurgical measurements, assuming matched samples with an α = .05 and 80% statistical power. Data are reported as mean (SD).
A total of 115 caregivers meeting the eligibility criteria were enrolled in the study; 14 were lost to follow-up and were thus excluded. Sampling bias was assessed to ensure that the enrolled population was representative of those eligible. Mean preoperative domain and survey scores, as well as physical examination variables, were compared between patients who completed the postoperative survey and those who did not. There were no differences in demographics, physical examination variables, or mean survey scores from visits 1 and 2. Demographic data, physical examination findings, and mean survey and domain scores were also compared by site and were not significantly different. Patient age was 6.2 (2.5) years. Forty-six percent of patients were girls and 54% were boys. Forty-two patients (41.6%) were from the northeastern United States, 27 (26.7%) were from the south, 19 (18.8%) were from the west, and 13 (12.9%) were from the central part of the country.
Sleep symptoms were documented by the examining attending physician (Table 1). All patients from whom data were obtained were snorers, and 91.8% had restless sleep. Only a small percentage of children had enuresis (13.4%), nocturnal sweating (7.2%), or morning headaches (3.1%). Physical characteristics, including tonsil size (Table 2) and OCF characteristics (Table 3), were also rated by the physician. Tonsil size was 3.4 (0.65). Orocraniofacial scores were tabulated and then divided into low, medium, and high scores by 33rd percentiles. The scores were 7.7 (1.3) in the lowest-scoring group (n = 37), 10.5 (0.51) in the mid-scoring group (n = 27), and 13.6 (1.5) in the high-scoring group (n = 35).
In 55 (54.5%) of 101 patients, diagnosis was based on history and physical examination alone; an additional diagnostic test was used in the remainder. Nasopharyngoscopy was performed in 10 patients (9.9%), and lateral neck films were taken in 6 (5.9%). Sleep audiotapes were obtained in 25 patients (24.8%); data from these audiotapes were not objectively rated. Sleep studies were obtained in 8 patients (7.9%). Of these 8 patients, the severity of obstructive sleep apnea was graded as mild in 50% and moderate in 50%. A respiratory distress index was available for 6 of these patients (mean, 6.7 [3.1]).
All caregivers easily self-administered the questionnaire in several minutes after a brief explanation of its contents. The distribution of baseline domain scores is shown in Figure 2. Physical suffering, sleep disturbance, and speech and swallowing difficulties were classified as moderate or greater problems (score of ≥3) for 82.1%, 95.0%, and 62.4% of patients, respectively; 52.5% of caregivers rated their level of concern over their child's sleep disturbance in the 2 most severe categories ("very much" and "could not be worse"), and 77% rated their level of concern as a moderate or worse problem. Median baseline survey responses for the 101 caregivers are given in Table 4. Sleep disturbance (score, 4.6 [1.1]), physical suffering (score, 3.9 [1.3]), and caregiver concern (score, 3.8 [1.9]) were the highest-rated items, whereas activity limitations (score, 1.5 [1.6]) were rated as the least affected.
At the initial visit, parents also completed the Satisfaction With Decision Scale and the Satisfaction With Office Visit Scale, validated measures of outpatient satisfaction. Overall, parents were satisfied with their decision to pursue surgery for their child's OSDs (Table 5) and were very satisfied with their office visit experience (Table 6).
Changes in QOL were determined by measuring the differences in scores between consecutive visits. Preoperative change in QOL was determined by the change in scores between visits 1 and 2, and postoperative change in QOL was determined by the change in score between visits 2 and 3. Patients were determined to have a valid change score if the time between surgery (visit 2) and the postoperative visit (visit 3) was 21 days or longer. This prevented sampling of symptoms in the recovery period, which could distort QOL ratings. Ninety-eight (97.0%) of 101 patients were determined to have valid change scores. Time between visits 1 and 2 was 31.0 (25.9) days (range, 4-162 days) and between visits 2 and 3 was 35.7 (14.8) days (range, 21-98 days) for these patients.
The change score before surgery (visit 1 to visit 2) was 0.18 (95% CI, 0-37), indicating trivial change overall. This change was not significantly different than zero (t97 = 1.91; P = .06). Before surgery, 42 (42.9%) of 98 patients had trivial or small changes in QOL, whereas 16 (16.3%) had moderate or large changes in QOL. Forty patients (40.8%) had worsening of their QOL before surgery. The change score after surgery (visit 2 to visit 3) was 2.3 (95% CI, 2.1-2.6), which indicates a large improvement in QOL (t97 = 15.7; P<.001). After surgery, 73 patients (74.5%) demonstrated a large degree of improvement in QOL (change score ≥1.5), 6 (6.1%) had a moderate improvement, 7 (7.1%) had a small improvement, and 7 (7.1%) had a trivial improvement. In addition, 5 patients (5.1%) had a worsening of their QOL after surgery: 1 (1.0%) to a large extent, 2 (2.0%) to a moderate extent, and 2 (2.0%) to a small extent. The differences in degree of clinical change before and after surgery were significant (χ24 = 87.2; P<.001).
Change scores were converted into a measure of effect size, the SRM, to facilitate interpretation of results and comparisons with other health status measures. Effect sizes before surgery indicated that effects were trivial or small for all domains and the survey as a whole (Table 7). After surgery, however, effect sizes were large for the mean survey score and all domains except activity limitations, which had a moderate effect. Excluding emotional distress and activity limitations, postoperative effect sizes greatly exceeded the 0.8 cutoff value for large change.11 The largest improvements were seen in physical suffering, sleep disturbance, and caregiver concern.
Multivariate analysis did not reveal any predictive factors for improvement in QOL other than undergoing adenotonsillectomy. Specifically, factors not related to outcome included age, sex, degree of tonsillar obstruction, OCF score, parental satisfaction with decision for surgery, parental satisfaction with office visit, and sleep symptoms. Tonsil obstruction greater than 75% was the only factor predictive of poorer QOL before surgery (P = .004).
We performed a variety of statistical tests to discern whether longer follow-up was related to change in QOL score because some prospective studies have indicated that QOL scores may change after surgery. Although follow-up ranged from 21 to 98 days, mean follow-up was 35.7 (14.8) days (median, 30 days). Note that 75% of patients were followed for 41 days or less; 90% were followed for 55 days or less. There were only 3 extreme outliers: 1 patient at 76 days and 1 each at 97 and 98 days. Linear regression analysis measuring any linear relationship between length of follow-up and improved QOL found no such relationship (r = 0.13; P = .21). In addition, using a standard t test, we divided the group into 2 groups: (1) 30 days or less of follow-up and (2) more than 30 days of follow-up. The difference in survey change score was statistically significant, with a change of 1.97 (1.6) (n = 50) vs 2.74 (1.24) (n = 48) for the 2 groups, respectively (P<.01). However, this 0.77 difference represents a small difference in QOL change score as defined by Juniper et al.10 We also calculated the rate of change of QOL change scores by taking the change scores and dividing by the length of follow-up. This calculation yields a "QOL improvement score per patient-month." We then compared the average rates for the 2 groups and found that the mean rate of improvement was slightly higher in the group with 30 days or less of follow-up compared with the group with more than 30 days of follow-up: 2.24 (1.8) vs 1.97 (0.99); however, this difference was not statistically significant (P = .35).
As decisions concerning resource allocation become increasingly stringent, it is important to understand the personal impact of diseases and their treatments beyond the standard medical morbidity or functional limitations so that this can be incorporated into the decision-making process. Procedures or treatments that have an impact on patient functioning above and beyond laboratory value improvements should be recognized. This study is the first, to our knowledge, to define QOL changes in children who undergo adenotonsillectomy for an OSD. We hope that this information will be helpful for clinicians to provide to parents who are considering the procedure for their child. Studies such as this will provide evidence for the effectiveness of this procedure to others involved in medical decision making.
General health status instruments enable comparisons of health status impact between diseases. A recent study using the Child Health Questionnaire, a general instrument of child health-related QOL, reported that the impact of adenotonsillar disease (recurrent infection and adenotonsillar hypertrophy) on child QOL is similar to that of juvenile rheumatoid arthritis.12 Disease-specific instruments such as the OSD-6 and others12-14 focus on areas of function that pertain to a particular disease or condition and are used to describe the impact of disease on individuals above and beyond the usual biomedical variables used to assess severity of the particular disease or condition.
The OSD-6 is a disease-specific instrument that has been shown to be of use as an evaluative measure, showing changes in QOL before and after an intervention (eg, adenotonsillectomy). In the present study, almost 90% of all children undergoing the procedure had improvements in QOL after surgery. Seventy-five percent had large improvements, and an additional 6% had moderate improvements in child QOL. Findings of this magnitude indicate the benefit of adenotonsillectomy for most children with OSDs. These findings support the objective improvements that are seen in follow-up polysomnographic studies and symptom questionnaires2-4,15 and quantify the subjective improvements that occur after the procedure.
As is standard in most outcome studies, this study was conducted using information obtained in routine clinical practice. Sleep studies were not obtained routinely and in fact were performed on only 8% of patients. This finding concurs with usual clinical practice because most adenotonsillectomies are performed without preoperative polysomnographic testing unless there are special circumstances.16,17 Sleep studies that are available to most practitioners are not considered diagnostic of the spectrum of OSDs because of the lack of esophageal pressure monitoring and the subsequent inability to diagnose UARS.1,9,18,19 Inability to diagnose UARS would lead to the misdiagnosis of children with significant sleep fragmentation.1 A recent validation study14 of an OSD survey showed a statistically significant but only fair correlation with respiratory distress indices obtained from nap studies. This finding is likely because (1) a significant proportion of children with problematic sleep-disordered breathing have normal sleep study findings and (2) nap studies have been validated for use in adults only,20 and their sensitivity in detecting sleep-disordered breathing in children is so far undemonstrated, to our knowledge. Clearly, accurate diagnosis of the full spectrum of OSDs in children is still in evolution. However, as stated previously by others21 in reference to children with "normal" polysomnograms who underwent adenotonsillectomy subsequently, "otolaryngologists with good clinical sense decided to pursue the appropriate course of action despite a normal test." As the diagnostic process evolves for determining who will benefit from surgery in a readily available and reliable manner, it is helpful to know that most children with compelling history and physical examination findings of OSDs who are selected for surgery will have large improvements in their QOL. In fact, given that 75% of children had large improvements and 88% had some degree of improvement in QOL after adenotonsillectomy, the debate over the necessity of routine polysomnographic testing may be moot.
Regarding the 5 children who worsened after surgery, no predictive factors could be demonstrated, although the small sample size of patients in this category would limit the ability to find significant trends. Four of 5 patients who worsened were seen after surgery at 22 days, so it is possible that they were still experiencing symptoms of recovery. We do not have longer-term data to know whether these patients ultimately improved. The fact that a small percentage of patients seems to worsen after surgery without explanation merits discussion with the parents, as does any other possible outcome discussed in obtaining informed consent.
The times at which patients were seen for postoperative follow-up were variable (range, 21-98 days) because of factors beyond our control. However, mean follow-up was 35.7 days and median follow-up was 30 days. Note that 75% of patients were followed for 41 days or less. Ninety percent of patients were followed for 55 days or less. Linear regression analysis found no relationship between length of follow-up and improved QOL. We divided the group into 2 subgroups based on length of follow-up: (1) 30 days or less of follow-up and (2) more than 30 days of follow-up. The difference in survey change score was statistically significant, with a change of 1.97 (1.6) (n = 50) vs 2.74 (1.24) (n = 48) (P<.01) for the 2 groups, respectively, using a standard t test. This 0.77 difference represents a small improvement in QOL change score as defined by Juniper et al.10 This analysis reflects that QOL may continue to improve even beyond the early postoperative period (≤30 days), which would be a desirable outcome for these patients. This finding makes sense considering the usual clinical course of patients who demonstrate overall improved health and well-being within several months of adenotonsillectomy. We do not have longer-term follow-up data on these patients to be able to comment on whether this finding holds true for these patients.
As seen in the previous study,6 there was poor correlation between clinical ratings of degree of airway obstruction and the mean survey score. In the present study, we attempted to find a better model for defining physical characteristics that would be predictive of improvement in QOL. Guilleminault et al9 showed that high scores on the clinical OCF scale and tonsillar scale were correlated with a diagnosis of UARS and obstructive sleep apnea in 411 children presenting to a sleep center for evaluation. In the present study, these factors were not predictive of improved QOL in multivariate analysis. However, virtually all of our patients had a degree of tonsillar obstruction greater than 50%, so the data may have been too homogeneous to give a meaningful correlation. Regarding the OCF score, we found that children in all categories (low, medium, or high) had substantial improvements in QOL. This may indicate a lack of sensitivity of the scale or incorrect categorization of the physical characteristics by the many physicians participating in the study. It could also reflect that children with all types of OCF characteristics can achieve a large benefit in QOL from adenotonsillectomy. A larger sample size is needed to further clarify these findings.
The results of the parental Satisfaction With Decision and Satisfaction With Office Visit scales were also not correlated with patient outcome. This was an important finding because we wanted to show that improvements after surgery were due to the surgery itself and were not biased by parents who were exceptionally satisfied with the perioperative care. Findings from the satisfaction survey were similar to those of Rosenfeld et al7 in parents who were having their children evaluated for tympanostomy tube placement. Sixty-five percent of parents rated the overall visit as excellent, and 93% rated their visit as very good or excellent. The lowest area of satisfaction was concerning the explanation by the physician of what was done for the child. This was rated as excellent by only 69% of parents, although it was rated as very good and good by an additional 27% and 4% of parents, respectively. This finding reiterates the need to make sure parents have a true understanding of visit proceedings. Findings on the Satisfaction With Decision Scale were also independent of outcome. Results showed that only about two thirds of parents felt strongly that they were told enough to make a good decision for their child, felt it was the best decision for their child, expected to follow through with the decision, and were content overall with the decision to have surgery. These findings may indicate the apprehension that parents feel when electing to have a surgical procedure performed on their child, but they may also reinforce the need for adequate physician education so that parents can be as comfortable as possible with their decision.
Strengths of this study include the high generalizability of the findings given the geographically diverse regions in which the study was performed, which included urban and suburban populations. The findings of this study are also easily generalizable to clinical practice because patients included in this study likely represent children all along the spectrum of sleep-disordered breathing. No effort was made to limit study participants to those with documented polysomnographic abnormalities. As discussed previously, this could have led to the exclusion of children with significant sleep fragmentation.
Another limitation is the short time after surgery at which follow-up questionnaires were obtained. Most postoperative visits were scheduled beyond the point (4-5 weeks) at which many clinicians typically see their patients (2-3 weeks) to avoid sampling symptoms of recovery. We found that compliance with the postoperative visit is often poor, and we believed that extension of the time any later after surgery would decrease patient return rates even further. Therefore, our results can only give insight into short-term QOL improvements. There is no reason to believe that the OSD would recur in most children, although there is a small possibility of symptomatic adenoid regrowth.22 In addition, there is the possibility that the onset of puberty in boys can lead to upper airway anatomic changes that can cause recurrence of sleep-disordered breathing.1,23,24 Further research is needed to comment on the long-term effects of adenotonsillectomy on child QOL. However, as more time elapses after surgery for patient evaluation, the more difficult it will be to dissect out improvement in QOL due to the surgical procedure from other medical, social, and developmental factors.
In conclusion, most children with OSDs who are selected for surgery in pediatric otolaryngology practices will have a large improvement in QOL, at least in the short term, after adenotonsillectomy.
Accepted for publication October 26, 2001.
This study was presented at the American Society of Pediatric Otolaryngology meeting, Scottsdale, Ariz, May 10, 2001.
We thank Farrel J. Buchinsky, MD, and Jeffrey Keller, MD, for contributing patient data to this study and Leslie Schmidt, RN, Anna Squillante, RN, Amelia Morris, RN, Sandra Neal, RN, Monica Belmonte, MD, and Lynn Golembiewski, RN, for their help with data collection.
Corresponding author and reprints: Lianne M. de Serres, MD, Division of Pediatric Otolaryngology, The Children's Hospital of New York, 3959 Broadway, Room 501N, New York, NY 10032 (e-mail: LMD54@columbia.edu).